A hybrid layer-multiplexing and pipeline architecture for efficient FPGA-based multilayer neural network

スポンサーリンク

概要

• 論文の詳細を見る

• This paper presents a novel architecture for an FPGA-based implementation of multilayer Artificial Neural Network (ANN), which integrates both the layer-multiplexing and pipeline architecture. Given a kind of FPGA to be used, the proposed method aims at enhancing the efficiency of resource usage of the FPGA and improving the forward speed at the module level, so that a larger ANN can be implemented on traditional FPGAs and also a high performance is achieved. Usually FPGA board is not changed for every applications, thus, we need not mind about the usage of it if the application can be implemented within the resource limitation. We developed a new mapping method from ANN schematic to FPGA by using this hybrid architecture, and also developed an algorithm to automatically determine the architecture by optimizing the application specific neural network topology. The experimental results show that the proposed architecture can produce a very compact circuit for multilayer ANN to meet resource limitation of a given FPGA. Furthermore, higher performance is obtained as compared with conventional methods.

著者

• Li Ce

  Graduate School Of Information Production And Systems Waseda University

• Dong Yiping

  Graduate School Of Information Production And Systems Waseda University

• Watanabe Takahiro

  Graduate School of Information, Production and System, Waseda University

• WATANABE TAKAHIRO

  Graduate School of Environmental Studies, Tohoku University

• Lin Zhen

  Graduate School of Information, Production and System, Waseda University

関連論文

• Circuit Design Optimization Using Genetic Algorithm with Parameterized Uniform Crossover
• Score Sequence Pair Problems of (r_, r_, r_)-Tournaments : Determination of Realizability(Graph Algorithms,Foundations of Computer Science)
• Generation of high-density atto-second electron bunch by intense short pulse laser (特集 平成18年基礎・材料・共通部門大会)
• Mixed Constrained Image Filter Design for Salt-and-Pepper Noise Reduction Using Genetic Algorithm (特集 非線形電子回路)
• Analysis before Starting an Access : A New Power-Efficient Instruction Fetch Mechanism
• DS-1-3 DESIGN AND IMPLEMENTATION OF SHA-1 HASH FUNCTION USING VERILOG HDL
• Region-Oriented Placement Algorithm for Coarse-Grained Power-Gating FPGA Architecture
• Low Power Placement and Routing for the Coarse-Grained Power Gating FPGA Architecture
• A New Recovery Mechanism in Superscalar Microprocessors by Recovering Critical Misprediction
• A hybrid layer-multiplexing and pipeline architecture for efficient FPGA-based multilayer neural network
• A Behavior-based Adaptive Access-mode for Low-power Set-associative Caches in Embedded Systems
• A Behavior-based Adaptive Access-mode for Low-power Set-associative Caches in Embedded Systems
• Distribution of artificial radionuclides (^Ag, ^Te, ^Cs, ^Cs) in surface soils from Miyagi Prefecture, northeast Japan, following the 2011 Fukushima Dai-ichi nuclear power plant accident
• An Efficient Algorithm for 3D NoC Architecture Optimization
• An Efficient Algorithm for 3D NoC Architecture Optimization
• A Sophisticated Routing Algorithm in 3D NoC with Fixed TSVs for Low Energy and Latency

もっと見る 閉じる

スポンサーリンク